Force field based molecular structure and conformer generation

Abstract

Systems and methods for molecular structure generation and conformer elaboration, in which natural physical molecular movements can be combined with a molecular force field that constrains those movements, to rapidly produce conformational variants with relatively low energy. Construction and energy minimization of initial and subsequent 3D molecular models using force field parameters, are combined with alteration of the molecular model using biophysical transformations, to generate one or more conformations thereof. The biophysical transformations each include natural physical movements of parts of the molecule, such as rotation of atoms or bonds about a selected axis in the 3D molecular model. The selected axis can define ring components, selected bonds within a macrocyclic ring, selected bonds joining substituents or other portions of the molecule, or axes or lines defining one or more geometric features of the molecular model. Once altered using biophysical transformations, the 3D molecular model can also have energy minimization performed with respect to a molecular force field model. A subset of the generated conformers can be collected, compressed from time to time, and selected for output.

The invention claimed is: 
     
       1. A method for generating conformational variations of three-dimensional (3D) molecular structures of organic molecules, the method including steps of
 receiving one or more representations of a selected molecule, said representations identifying assignment of 3D atomic positions and atomic bonds; 
 identifying one or more ring systems in the selected molecule; 
 performing one or more biophysical transformations on the ring systems, said biophysical transformations representing natural physical adjustment of one or more of the atomic positions with respect to a geometric or topological feature of the selected molecule; 
 selecting one or more conformational variants in response to a measure of molecular energy; and 
 repeating the steps of performing, producing, and selecting until a termination condition is reached; 
 wherein the steps of selection of conformational variants are responsive in multiple stages to an energy window above a selected currently known minimum energy conformational variant, with the energy window having larger windows than later stages, and with the energy window for a final stage being responsive to a specified parameter; and 
 wherein the steps of selection of conformational variants are responsive to a maximum number of individual conformers. 
 
     
     
       2. A method as in  claim 1 ,
 including steps of 
 torsional sampling for rotatable bonds outside of ring systems; 
 selection of diverse conformational variants; or 
 energy minimization with respect to a force field. 
 
     
     
       3. A method as in  claim 1 ,
 wherein the terminating condition includes one or more of: 
 a maximum number of conformational variants are identified, the maximum number of conformational variants being responsive to one or more of: a number of rotatable bonds in the molecule, a user specification; 
 a maximum number of attempts to identify conformational variants are made; 
 a maximum amount of computation or run time is used. 
 
     
     
       4. A method as in  claim 1 ,
 wherein the steps of selecting are responsive to one or more of: 
 a measure of differences between conformational variants, 
 an evaluation of configurational correctness. 
 
     
     
       5. A method for generating conformational variations of three-dimensional (3D) molecular structures of organic molecules, the method including steps of
 receiving one or more representations of a selected molecule, said representations identifying assignment of 3D atomic positions and atomic bonds; 
 identifying one or more ring systems in the selected molecule; 
 performing one or more biophysical transformations on the ring systems, said biophysical transformations representing natural physical adjustment of one or more of the atomic positions with respect to a geometric or topological feature of the selected molecule; 
 selecting one or more conformational variants in response to a measure of molecular energy; and 
 repeating the steps of performing, producing, and selecting until a termination condition is reached; 
 wherein the steps of selecting are responsive to 
 eliminating redundant conformational variants from ring system elaboration in response to thresholds on RMS deviation that depend on ring system size or upon user specification; 
 eliminating redundant conformational variants from torsional elaboration in response to thresholds on RMS deviation that depend upon user specification; or 
 compression of conformational pools in response to selection of maximally different conformational variants. 
 
     
     
       6. A method as in  claim 5 ,
 including steps of 
 torsional sampling for rotatable bonds outside of ring systems; 
 selection of diverse conformational variants; and 
 energy minimization with respect to a force field. 
 
     
     
       7. A method as in  claim 5 ,
 wherein the terminating condition includes one or more of: 
 a maximum number of conformational variants are identified, the maximum number of conformational variants being responsive to one or more of: a number of rotatable bonds in the molecule, a user specification; 
 a maximum number of attempts to identify conformational variants are made; 
 a maximum amount of computation or run time is used. 
 
     
     
       8. A method for generating conformational variations of three-dimensional (3D) molecular structures of organic molecules, the method including steps of
 receiving one or more representations of a selected molecule, said representations identifying assignment of 3D atomic positions and atomic bonds; 
 identifying one or more ring systems in the selected molecule; 
 performing one or more biophysical transformations on the ring systems, said biophysical transformations representing natural physical adjustment of one or more of the atomic positions with respect to a geometric or topological feature of the selected molecule; 
 selecting one or more conformational variants in response to a measure of molecular energy; and 
 repeating the steps of performing, producing, and selecting until a termination condition is reached; 
 wherein the ring systems include at least one macrocyclic structure having a connected ring structure of size nine or more, 
 wherein the macrocyclic structure includes a torsional group of four atoms, the central two having a bond defining a twisting axis, a third atom defining an anchor, and a fourth atom defining a twistable atom; 
 wherein the biophysical transformation includes a ring twist, the ring twist including rotating the twistable atom relative to the twisting axis, while the anchor atom maintains its position relative to the twisting axis; 
 wherein the biophysical transformation includes performing the ring twist and conducting energy minimization with respect to a force field. 
 
     
     
       9. A method as in  claim 8 ,
 wherein pinning the ring twist is of sufficient magnitude to prevent reversion or lack of sampling. 
 
     
     
       10. A method as in  claim 8 ,
 wherein pinning the ring twist includes a square-welled quadratic penalty associated with a freedom of movement for the three atoms being held as fixed. 
 
     
     
       11. A method as in  claim 8 ,
 wherein ring twists include increments of regular increments, including 180 degrees and a rotation beyond 180 degrees. 
 
     
     
       12. A method as in  claim 8 ,
 wherein the steps of conducting energy minimization with respect to a force field 
 include steps of 
 conducting energy minimization while pinning the anchor atom, the twistable bond, and the twistable atom; 
 relaxing the pin; and 
 conducting energy minimization without pinning the anchor atom, the twistable bond, and the twistable atom. 
 
     
     
       13. A method as in  claim 8 ,
 wherein the steps of performing the ring twist are performed for each ring twist in the macrocyclic structure. 
 
     
     
       14. A method as in  claim 8 ,
 wherein the steps of performing the ring twist are performed for more than one amount of rotation about the twistable bond. 
 
     
     
       15. A method as in  claim 8 ,
 including steps of 
 torsional sampling for rotatable bonds outside of ring systems; 
 selection of diverse conformational variants; and 
 energy minimization with respect to a force field. 
 
     
     
       16. A method as in  claim 8 ,
 wherein the terminating condition includes one or more of: 
 a maximum number of conformational variants are identified, the maximum number of conformational variants being responsive to one or more of: a number of rotatable bonds in the molecule, a user specification; 
 a maximum number of attempts to identify conformational variants are made; 
 a maximum amount of computation or run time is used. 
 
     
     
       17. A method for generating three-dimensional (3D) molecular structures for conformational candidates of organic molecules, the method including steps of
 assigning initial atomic coordinates in response to molecular connectivity; 
 refining those coordinates to minimize molecular energy while adhering to constraints upon molecular configuration; 
 selecting candidate molecular structures in response to a measure of molecular energy and an assessment of congruence with specified molecular configuration; and 
 repeating the steps of assigning initial atomic coordinates, refining those coordinates, and selecting candidate molecular structures, until a termination condition is reached; 
 wherein the steps of assigning initial atomic coordinates include steps of 
 computing initial approximate atomic positions by modeling each atom as a tetrahedron and by using approximate bond lengths and torsional choices to avoid energetically unfavorable configurations; 
 wherein the steps of refining those coordinates include steps of 
 assigning initial equal atomic partial charges, 
 refining the initial positions using force field minimization including force field terms that include bond length, bond angle, and electrostatic terms, 
 imposing force field terms to enforce specified chirality at tetrahedral centers, followed by energy minimization, 
 imposing force field terms to enforce specified double-bond configurations along with all torsional terms, followed by energy minimization, and 
 removing special configurational force field terms, assigning partial charges using a standard method, and minimizing the energy using all standard force field terms; 
 repeating the preceding steps with alternative initial atomic coordinate assignments; and 
 wherein the steps of selecting candidate molecular structures include steps of 
 collecting the conformational variants so generated and evaluating each for configurational correctness and satisfaction of a maximum energy cutoff, 
 terminating the repetition of steps when a fixed number of conformers meet the evaluation criteria or when a maximal number of repetitions has occurred, and 
 producing the atomic coordinates of the conformer with lowest energy that passed the configurational correctness criteria. 
 
     
     
       18. A method as in  claim 17 ,
 including steps of 
 torsional sampling for rotatable bonds outside of ring systems; 
 selection of diverse conformational variants; and 
 energy minimization with respect to a force field.